MetaLearner Forecasting Stack White Paper

Take a look to our white paper, where we explore the various machine learning and mathematical models powering our platform.

In today's fast-paced business world, accurate forecasting can be the difference between leading the market and falling behind. At MetaLearner, we've developed an innovative approach to forecasting, leveraging cutting-edge AI and data science tools to help businesses stay ahead of trends and make smarter decisions.

Take a look to our white paper, where we explore the various machine learning and mathematical models powering our platform, ensuring unparalleled accuracy and reliability in forecasting.

A crucial task in timeseries forecasting is the early identification of the most suitable forecasting method. For the past decade we have been designing, coding and optimizing a general framework for forecast-model selection using meta-learning. An ensemble method approach is used within the ERP domain (SAP, Oracle and Snowflake) to identify the best forecasting method using time series features of the time series to select the class of models, or even the specific model, to be used for forecasting.

The framework is currently deployed with clients’ structured and unstructured data and is shown to yield accurate forecasts comparable to several benchmarks and other commonly used approaches of time series forecasting. A key advantage of our technology stack is that the time consuming process of building a classifier is handled in advance of the forecasting task by multiple AI agents. Please use the following link to access our latest 2 minutes video which shows how MetaLearner empowers users with our ERP Copilot: https://www.youtube.com/watch?v=xf9H5hQS5rM. It features seamless integration, data-driven insights, and democratized data science—all without the need for technical expertise. It enables end users to gain insights into their data and perform complex data operations through natural conversations.

All output produced by the AI will have sources attributed to it (SQL statements if queried from ERP databases, in-text citations if queried from a vector database). Additionally, we have integrated data science tools that allow the AI to make forecasts based on the data it receives, which is useful in scenarios such as demand planning and inventory optimization. To achieve this, we have created proprietary tools and functions for the AI to work with and have achieved great preliminary results thus far.

Last but not least, as an aspiring AI data science startup already working with enterprise customers, we have prioritized cybersecurity and data privacy. Leveraging existing technology, such as NVIDIA’s NeMo Guardrails, has allowed us to enhance the security of our AI systems. We not only isolate every software forecasting module to ensure security but also provide references and sources for how our AI derives each piece of information. Through a tightly woven architecture, we can minimize AI cybersecurity risks and instill user confidence in this burgeoning technological revolution.